CN109992359A - A kind of transaction scheduling method - Google Patents

A kind of transaction scheduling method Download PDF

Info

Publication number
CN109992359A
CN109992359A CN201910243782.4A CN201910243782A CN109992359A CN 109992359 A CN109992359 A CN 109992359A CN 201910243782 A CN201910243782 A CN 201910243782A CN 109992359 A CN109992359 A CN 109992359A
Authority
CN
China
Prior art keywords
affairs
thread
transaction
scheduling
scheduling unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201910243782.4A
Other languages
Chinese (zh)
Inventor
罗侍田
程招招
古小健
王建忠
陈家明
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Chuanglian Time Technology Co Ltd
Original Assignee
Shenzhen Chuanglian Time Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Chuanglian Time Technology Co Ltd filed Critical Shenzhen Chuanglian Time Technology Co Ltd
Priority to CN201910243782.4A priority Critical patent/CN109992359A/en
Publication of CN109992359A publication Critical patent/CN109992359A/en
Pending legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/466Transaction processing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/48Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806Task transfer initiation or dispatching
    • G06F9/4843Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/485Task life-cycle, e.g. stopping, restarting, resuming execution
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/54Interprogram communication
    • G06F9/546Message passing systems or structures, e.g. queues
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00Indexing scheme relating to G06F9/00
    • G06F2209/54Indexing scheme relating to G06F9/54
    • G06F2209/543Local
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2209/00Indexing scheme relating to G06F9/00
    • G06F2209/54Indexing scheme relating to G06F9/54
    • G06F2209/548Queue

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention relates to a kind of transaction scheduling methods, the scheduling unit of operating system handles affairs, the affairs of processing are divided into different transaction types, data structure is established for affairs to be treated, the data structure is connected into a queue, treatment process includes the following steps: step 1, pop-up affairs;Affairs are divided into a ready transaction queues and several synchronized transaction queues by step 2, scheduling unit;Step 3, scheduling unit choose an affairs according to rule from queue, distribute a thread for the affairs;Step 4, scheduling unit are assigned to affairs in CPU and handle.The beneficial effects of the present invention are: can effectively avoid mistake, while software bug is reduced, greatly improves software development efficiency.

Description

A kind of transaction scheduling method
Technical field
The present invention relates to a kind of transaction scheduling method of computer operating system, in particular to a kind of embedded computer behaviour Make the transaction scheduling method of system.
Background technique
It is using thread as executable unit, the operation of thread stops when current computer is run under the control of an operating system Only, the processes such as switching, obstruction, suspend mode, directly represent the operational process of CPU.Legacy operating system is as starting point, design A set of scheduling system, thread form several queues in memory.Operating system nucleus is then according to certain algorithm, CPU Time distributes to each thread.And a series of api function is provided, programmer calls these api functions, so that it may create line Journey, and the processes such as operation, stopping, suspend mode, obstruction for controlling thread.This process is the operational process height with computer It is relevant, fully belong to proceduring, structuring programming.
In modern software practices, the design of proceduring, structured software is had already fallen behind, more advanced design method, be towards Object programming.Some advanced software design tools, such as c# add on procedure-oriented and the operating system of structuring One layer of packaging, provides the exploitation environment of a set of object-oriented, does not need programmer in face of thread these procedural members Element.However, these tools can only use in PC, cell phone software exploitation, in the exploitation of embedded system, C language can only be still used Speech, can only directly control thread, and use is proceduring, structuring method develops software.
Summary of the invention
The purpose of the present invention is to propose to a kind of scheme of transaction scheduling, what is saved in calculator memory is to require computer complete At transaction queues, for scheduling unit according to dispatching algorithm, which affairs determination currently needs to handle, and CPU is just distributed to it.Behaviour Make system and a series of API is provided, programmer uses these API, so that it may which the treatment process for popping up affairs, controlling affairs is It is achieved through the following technical solutions.
The scheduling unit of a kind of transaction scheduling method, operating system handles affairs, the affairs of processing is divided into different Transaction types establish data structure for affairs to be treated, the data structure are connected into a queue, treatment process packet Include following steps:
Step 1, pop-up affairs;
Affairs are divided into a ready transaction queues and several synchronized transaction queues by step 2, scheduling unit;
Step 3, scheduling unit choose an affairs according to rule from queue, distribute a thread for the affairs;
Step 4, scheduling unit are assigned to affairs in CPU and handle.
Further, operating system is newly-built or distributes thread process affairs, and the attribute of thread must be corresponding with transaction types, The affairs of same type use identical thread process, and different types of affairs use different thread process.
Further, the corresponding thread of each affairs, if there is a plurality of same type of affairs need to handle, operation System can create multiple identical threads and handle simultaneously, or sequentially be handled with the same thread according to the feature of affairs.
In step 3, the affairs then create a thread without suitable thread for the affairs.
Further, when popping up affairs, Priority flag is defined for affairs.
In step 1, scheduling unit is scheduled according to affairs Priority field, and when popping up affairs, new affairs have Gao You When first grade, scheduling unit is executed: immediately treating;Otherwise, scheduling unit executes when pop-up: thread is not distributed or create, until this Thread is just created or distributed when affairs should be processed.
Further, transaction types are defined, write issued transaction function for the affairs.
Further, when registering a transaction types, defining operation system creation is used to handle the thread of the type affairs Required two key parameters: the stack space that issued transaction entrance function and the function need.
Further, operating system is that affairs create or distribute existing thread, executes the corresponding affairs of the type affairs Handle function.
Further, after issued transaction is completed, operating system can recycle the occupied resource of the thread automatically, or delete Thread.
The beneficial effects of the present invention are:
Programmer is not required to manipulate thread, the needing computer disposal of the task need to be only divided into affairs one by one by programmer Type, and independent issued transaction function is write for various types of affairs, then it is registered in system ?.When a transaction occurs, discovery (detecting) as long as the programmer of the affairs tells operating system " certain type transaction has occurred ", Operating system is automatically created or distributes or wake up suitable thread and goes to handle the affairs, and created in person without programmer or Wake up corresponding thread.It can effectively avoid mistake, while reducing software bug, greatly improve software development efficiency.
Detailed description of the invention
Fig. 1 is transaction scheduling schematic diagram in the prior art.
Fig. 2 is the transaction scheduling schematic diagram of the specific embodiment of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, in the embodiment of the present invention device and calculation method carry out it is clear Chu is fully described by, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments. Based on the embodiments of the present invention, obtained by those of ordinary skill in the art without making creative efforts all Other embodiments shall fall within the protection scope of the present invention.
As shown in Fig. 2, a kind of transaction scheduling method, the scheduling unit of operating system handles affairs, and the affairs of processing are drawn It is divided into different transaction types, establishes data structure for affairs to be treated, the data structure is connected into a queue, Treatment process includes the following steps:
Step 1, pop-up affairs;
Affairs are divided into a ready transaction queues and several synchronized transaction queues by step 2, scheduling unit;
Step 3, scheduling unit choose an affairs according to rule from queue, distribute a thread for the affairs;
Step 4, scheduling unit are assigned to affairs in CPU and handle.
Specific embodiments of the present invention are by taking a kind of embedded system of entitled Dujiang weir (DJYOS) as an example, hereinafter referred to as originally Embodiment.
When popping up affairs, Priority flag is defined for affairs.In computer system, program can occur various each during running The affairs of sample, different types of affairs are handled by respective program, need to go to be distinguish with transaction types.Apply object-oriented Concept, transaction types are equivalent to the class of C++, and registration transaction types are equivalent to one class data type of statement;Affairs are equivalent to Object, pop-up affairs are equivalent to definition object, while doing the work of some constructed fuctions;Issued transaction completion is equivalent to revocation one A object, while doing the work of some similar destructor functions.
In specific embodiments of the present invention, the operational process of application program is exactly constantly to pop up new affairs and processing thing The process of business.Each affairs must belong to the transaction types being registered for, the affairs of same type using identical program into Row processing.
In specific embodiments of the present invention, the program for handling affairs is run in thread, thread as affairs resource and In the presence of, and the affairs are exactly the owner of thread, and therefore, any thread cannot all be born for no reason at all, exist and dead, It must be together with a certain business associates of a certain type.Thread is given birth to the needs of affairs, is completed and is withered away with affairs. Thread is without registration, and without having user to establish and starting, its creation, starting and deletion is automatically performed by operating system 's.This is different with legacy operating system, and legacy operating system can arbitrarily be created thread by user, and creation one is meaningless Thread be allowed.
Scheduling of the invention is according to being a ready transaction queues and several synchronized transaction queues, rather than thread queue (operating system having is also referred to as task queue), it is basic in the present invention just there is no thread queue.Scheduling of the invention is directed to thing Business rather than thread, creating a thread is because its owner needs processed, and it is because of the thread that thread, which is cut into, Owner needs to be cut into, and it is because its owner is suspended that thread, which is cut off,.Using thread as the positive of the resource of affairs Meaning is that, when some type of affairs frequently occur, operating system will assemble more resources, creates multiple threads for it to locate The affairs are managed, if these threads are distributed to different processors, processor itself in multiprocessor (multicore) system Just become a kind of resource, will greatly facilitate the management of multicomputer system.And in traditional programmed method, if programmer creates Main line journey is standby, and the corresponding one or several kinds of tasks of per thread wake up thread and handled after the generation of corresponding task, this Kind mode is more complex in terms of managing multiprocessor.Therefore, the present invention in a multi-processor system, has inborn superior Property.
The attribute of thread must be corresponding with transaction types, and the affairs of same type use identical thread process, inhomogeneity The affairs of type use different thread process.When one transaction types of user registration, it is necessary to which incoming operating system creation is for locating Manage two key parameters required for the thread of the type affairs: the stack space that issued transaction entrance function and the function need. After some type of affairs occur, operating system will create thread (or distributing existing thread) in due course and execute and be somebody's turn to do The corresponding issued transaction function of type transaction.After issued transaction is completed, it is occupied that operating system can recycle automatically the thread Resource also will be deleted thread when necessary.
Each affairs correspond to a thread, and if there is a plurality of same type of affairs need to handle, operating system can root It according to the feature of affairs, creates multiple identical threads and handles multiple identical affairs simultaneously, or sequentially located with the same thread Reason.This can more reasonably use computer resource.Although establishing multiple threads can't be than list in the case where uniprocessor A thread forcibly occupies processor for a long time and more makes full use of processor, but can produce the effect of multiple same type affairs parallel processings Fruit, such as multiple windows are drawn simultaneously, in particular, in multiprocessor (or multicore) system, it can be same what is frequently occurred Type transaction is assigned on different processors.And under legacy operating system, thread is created by programmer, if programmer A thread only is created for a certain work, then busy also only one thread works again for it for the work, the thread process The multi-task can only be executed serially.But creation thread may cause resource exhaustion repeatedly in this way, for example while handling certain affairs needs Serial ports is used, and serial ports is occupied by other threads, and the affairs occur in serial ports occupied period, operating system will be again Thread is created for it, which starts that blocked state can be entered because serial port resource is busy after executing, if affairs repeated, Operating system will create thread repeatedly for it, until ruing out of all memories, cause memory exhausted.Resource occurs in order to prevent Exhausted accident provides vpus_limit member in transaction types control block, indicates that the type affairs can establish line simultaneously The number of journey.
Specific embodiments of the present invention are to be scheduled based on affairs and the present invention and the maximum area of legacy operating system Not.Fig. 1 is abstracted from legacy operating system, and either simple OS or complexity OS, scheduling are all based on thread 's.In Fig. 1, thread either is created in initialization creation thread or thread implementation procedure, is all when creating thread Resource required for thread is distributed, stack is indispensable part therein.When popping up affairs, unless the priority of new affairs is sufficiently high, need It immediately treats, will not distribute or create thread when otherwise popping up, just created when the affairs should be processed or divide With thread.
Seem, other than it cannot create thread from interrupt processing function, two figures have very big similitude, in fact not So, is creation new thread very careful in thread implementation procedure why when programming under traditional OS? in order to create thread And cause to block, and what is be blocked is current thread.If the priority of new thread is lower than current thread, that current thread is just It kids, because of the thread of one low priority of creation, and high priority thread is caused to be blocked, this does not allow in RTOS, From scheme Design angle and unreasonable;Even if not being blocked, in high priority thread implementation procedure, when Hua great Liang Between for creating the thread of low priority and unreasonable.So all threads are always initial when programming under traditional OS The creation of change stage, seldom creates in the process of running.And the embodiment of the present invention is then without this problem, if the thing newly popped up Priority of being engaged in is not high enough, not can create thread for him.This more distribute computer in the present embodiment rationally Resource, it is assumed that have a product, there is uart serial ports, under traditional thread mode, uart communication thread be initialization when wound It builds, even if uart communication line does not connect, uart thread also must committed memory resource.If uart is communicated fairly frequently Occurring, that thread created when also can only be by initializing is bit by bit handled, even if you have multiple cpu cores, other cpu cores Also can only worry.Cannot that create multiple threads? can be with, Thread Pool Technology be exactly it is dry in this way, how many line created actually Journey is just suitable, is a very troubling problem.At traditional OS, Thread Pool Technology is a special subject.And In the transaction scheduling of the present embodiment, if not having data on uart communication line, the affairs are never popped up, would not also be accounted for With any resource.If uart mouthfuls of frequently communications, will frequently pop up affairs, if computer has multiple cpu cores, naturally Ground can be distributed to affairs on different cpu core, even if not knowing the programmer that thread is what, also can be carried out optimization Multicore programming.
Under traditional scheduling method based on thread, operating system only knows which thread is occupying CPU, cannot but know The road thread also can only distribute CPU for thread in What for, scheduler, cannot be for specific affairs handled by computer point With CPU.Therefore, under legacy operating system, programmer must skillfully grasp the knowledge in relation to thread, it is necessary to oneself be program needs The affairs of processing create thread, clear to have which thread to be currently running under which situation.And transaction scheduling is then different, Yong Huke Can not know thread presence and who be currently running, who is being waited, in fact, programmer does not need to be concerned about at all These.In the scheduling unit of the present embodiment, which affairs programmer only knows just processed, which affairs has been handled At which affairs is just medium to be processed in queue.It defines transaction types one by one and is registered in system, be every class affairs Processing function is write, is whole work of programming.When a plurality of affairs of same type are passing when having different priority A thread is established under system operating system or for each possible priority to realize or dynamically change in commission The priority of thread, regardless of which kind of mode, programmer requires to expend considerable time and effort, to ensure affairs by correct Priority is handled.And the present embodiment is different, it is exactly using affairs priority as the basis of scheduling, as long as generating that it is congenital When affairs, Priority flag is directly done in affairs can.Such as a serial communication program, interrupt function are responsible for bottom It receives, after receiving complete data packet, just issues upper level applications processing, upper level applications processing receives all Data packet, and have a command word domain in data packet, the priority of different order requests is different.Under legacy operating system, A comm_app thread is created, is changed while the thread is conveyed data in interrupt function according to command word The priority of comm_app thread, this way for changing thread priority in interrupt function are in many operating systems Do not allow;Another method is, for each corresponding possible priority of command word, to create an identical thread, These threads are in addition to priority is different, and other parts are identical, when interrupt function receives complete data packet, just according to number According to the command word in packet, message is sent out to the thread of respective priority.This method is although feasible, but can consume many CPU moneys Source, and be difficult in all possible priority of coding stage exhaustion, once command word changes, it is likely that it just needs to modify soft Part source code.And the scheduling unit of the present embodiment is different, programmer only needs to define a transaction types, and writes affairs for it It handles function proc_uart (), when interrupt function receives complete data packet, pops up directly corresponding with command word when affairs Priority can be scheduled as parameter, the scheduling unit of the present embodiment according to affairs Priority field automatically.
In addition, embedded system is mostly stress response system, the main task of stress response system is made correctly to extraneous affairs And timely react, from this view, programmer is not with process and thread these things are known, for processing external affairs Establishing thread is actually to do something against one's will, and under legacy operating system, you must do these work, your system could be correct A little things are done for you in ground.The present invention is based on the scheduling of affairs to be very suitable to this stress response system, and scheduled target is exactly to reflect The affairs of outside stimulus, rather than the thread of these outside stimulus is handled, meet the habitual thinking of people.Even if right and wrong stress System or non real-time system, the scheduling based on affairs still have its advantage, " busy just to do, to have nothing to do and take a seat " be people most For the mode of thinking of habit, obviously meet this mode of thinking by the scheduling strategy of thread of affairs, and is accustomed to people Property the identical scheduling mode of thinking, and be to avoid mistake, reduce the effective ways of software bug.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding And modification, the scope of the present invention is defined by the appended.

Claims (7)

1. a kind of transaction scheduling method, which is characterized in that the scheduling unit of operating system handles affairs, and the affairs of processing are divided For different transaction types, data structure is established for affairs to be treated, the data structure is connected into a queue, is located Reason process includes the following steps:
Step 1, pop-up affairs;
Affairs are divided into a ready transaction queues and several synchronized transaction queues by step 2, scheduling unit;
Step 3, scheduling unit choose an affairs according to rule from queue, distribute a thread for the affairs;
Step 4, scheduling unit are assigned to affairs in CPU and handle.
2. transaction scheduling method according to claim 1, which is characterized in that operating system is newly-built or distributes thread process thing Business, the attribute of thread must be corresponding with transaction types, and the affairs of same type use identical thread process, different types of thing It make sures with different thread process.
3. transaction scheduling method according to claim 2, which is characterized in that scheduling rule is: each affairs corresponding one A thread, if there is a plurality of same type of affairs need to handle, operating system can create multiple identical according to the feature of affairs Thread handle simultaneously, or sequentially handled with the same thread.
4. transaction scheduling method according to claim 1, which is characterized in that in step 3, the affairs are without suitable line Journey then creates a thread for the affairs.
5. transaction scheduling method according to claim 1, which is characterized in that when pop-up affairs, define priority for affairs Mark.
6. transaction scheduling method according to claim 4, which is characterized in that in step 1, scheduling unit is preferential according to affairs Grade domain is scheduled, and when popping up affairs, when new affairs have high priority, scheduling unit is executed: immediately treating;Otherwise, bullet Scheduling unit executes when out: not distributing or creates thread, just create or distribute thread when the affairs should be processed.
7. transaction scheduling method according to claim 3, which is characterized in that after issued transaction is completed, operating system meeting Automatically the occupied resource of the thread is recycled, or deletes thread.
CN201910243782.4A 2019-03-28 2019-03-28 A kind of transaction scheduling method Pending CN109992359A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910243782.4A CN109992359A (en) 2019-03-28 2019-03-28 A kind of transaction scheduling method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910243782.4A CN109992359A (en) 2019-03-28 2019-03-28 A kind of transaction scheduling method

Publications (1)

Publication Number Publication Date
CN109992359A true CN109992359A (en) 2019-07-09

Family

ID=67131009

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910243782.4A Pending CN109992359A (en) 2019-03-28 2019-03-28 A kind of transaction scheduling method

Country Status (1)

Country Link
CN (1) CN109992359A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110347482A (en) * 2019-07-18 2019-10-18 哈尔滨汇拓投资中心(有限合伙) OLTP affairs binding rule and queuing model improved method based on OLTPShare
CN112783613A (en) * 2019-11-07 2021-05-11 北京沃东天骏信息技术有限公司 Unit scheduling method and device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2004030663A (en) * 2002-06-20 2004-01-29 Internatl Business Mach Corp <Ibm> Device and method for integrating workload manager with system task scheduler
CN101770398A (en) * 2008-12-26 2010-07-07 罗侍田 Operating system kernel
CN106325980A (en) * 2015-06-30 2017-01-11 中国石油化工股份有限公司 Multi-thread concurrent system
CN109117260A (en) * 2018-08-30 2019-01-01 百度在线网络技术(北京)有限公司 A kind of method for scheduling task, device, equipment and medium

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2004030663A (en) * 2002-06-20 2004-01-29 Internatl Business Mach Corp <Ibm> Device and method for integrating workload manager with system task scheduler
CN101770398A (en) * 2008-12-26 2010-07-07 罗侍田 Operating system kernel
CN106325980A (en) * 2015-06-30 2017-01-11 中国石油化工股份有限公司 Multi-thread concurrent system
CN109117260A (en) * 2018-08-30 2019-01-01 百度在线网络技术(北京)有限公司 A kind of method for scheduling task, device, equipment and medium

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110347482A (en) * 2019-07-18 2019-10-18 哈尔滨汇拓投资中心(有限合伙) OLTP affairs binding rule and queuing model improved method based on OLTPShare
CN110347482B (en) * 2019-07-18 2021-07-23 哈尔滨汇拓投资中心(有限合伙) OLTP transaction combination rule and queue model improvement method based on OLTPShare
CN112783613A (en) * 2019-11-07 2021-05-11 北京沃东天骏信息技术有限公司 Unit scheduling method and device
CN112783613B (en) * 2019-11-07 2024-03-01 北京沃东天骏信息技术有限公司 Method and device for scheduling units

Similar Documents

Publication Publication Date Title
CN101226487B (en) Method for implementing inner core level thread library based on built-in Linux operating system
US7370326B2 (en) Prerequisite-based scheduler
JP2829078B2 (en) Process distribution method
US20230185607A1 (en) Hardware accelerated dynamic work creation on a graphics processing unit
CN100481014C (en) System and method for processing information
CN105808334B (en) A kind of short optimization of job system and method for MapReduce based on resource reuse
TWI736001B (en) Heterogeneous cpu/gpu systems for accelerating dataflow signal processing and method for accelerated dataflow signal processing performed by the same
Willhalm et al. Putting intel® threading building blocks to work
CN112463709A (en) Configurable heterogeneous artificial intelligence processor
CN112465129A (en) On-chip heterogeneous artificial intelligence processor
CN108595282A (en) A kind of implementation method of high concurrent message queue
CN107273141A (en) Embedded real-time operating system
CN110297661B (en) Parallel computing method, system and medium based on AMP framework DSP operating system
KR20200014378A (en) Job management
EP3877853A1 (en) Computing resource allocation
CN109240825A (en) Elastic method for scheduling task, device, equipment and computer readable storage medium
WO2021022964A1 (en) Task processing method, device, and computer-readable storage medium based on multi-core system
CN114579285B (en) Task running system and method and computing device
CN114138434B (en) Big data task scheduling system
CN109992359A (en) A kind of transaction scheduling method
CN103262035A (en) Device discovery and topology reporting in a combined CPU/GPU architecture system
CN100383743C (en) Real-time task scheduling method in Java operating system
CN101369258B (en) Input/output control system
US11301304B2 (en) Method and apparatus for managing kernel services in multi-core system
CN109766131A (en) The system and method for the intelligent automatic upgrading of software is realized based on multithreading

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20190709

RJ01 Rejection of invention patent application after publication